Re: [ESPResSo-users] About correlation property

[Peter Kosovan]

Hi Marc,

For Fickian diffusion msd=2dDt where d is the number of dimensions. 
square_distance_componentwise computes the square distance per each direction 
in space, so you have to substitute d=1. Note that the output of your 
correlation has 2+3*800 columns (if you are computing msd of all particles). 
The correlator does not know about the physical meaning of your observables. It 
treats them as vectors and applies the given operation. If you want to get an 
msd in 3D, you need to post-process the output.

One more remark for diffusion constant: if you're using Langevin thermostat and a 
low concentration of particles, the motion is ballistic until about tau=1/gamma. 
To get a good diffusion constant, you should only fit times tau>10/gamma.

Cheers,

peter

--
Peter Košovan
Institute of Computational Physics
University of Stuttgart
Pfaffenwaldring 27
70569 Stuttgart
Germany
www.icp.uni-stuttgart.de

On Tue, 6 Nov 2012, address@hidden wrote:

> Hi Everybody
>
> I am using the new correlation property to calculate diffusion constant in
> a system with 800 beads. But a question arise , the msd number isn't the
> diffusion constant by a factor of 6. This is right or 3D is taken in
> account an only a factor 2 applies ?
>
> Below is as I set the script.
>
> set pox [observable new particle_positions all]
>
> set msd [correlation new obs1 $pox corr_operation
> square_distance_componentwise tau_lin 2 tau_max 2 dt 0.005 compress1
> discard1]
>
> correlation $msd autoupdate start
>
> Best Regards
>
> Dr. Marc Segovia
> CCBG-DETEMA
> Facultad de Quimica
> Universidad del Republica
> 11700 Montevideo
> Uruguay